Justice Stevens, with whom Justice Souter joins, dissenting.

Prior to our decision in New York v. Belton, 453 U. S. 454 (1981), there was a widespread conflict among both federal and state courts over the question "whether, in the course of a search incident to the lawful custodial arrest of the occupants of an automobile, police may search inside the automobile after the arrestees are no longer in it." Id., at 459. In answering that question, the Court expanded the authority of the police in two important respects. It allowed the police to conduct a broader search than our decision in Chimel v. California, 395 U. S. 752, 762-763 (1969), would have permitted,1 and it authorized them to open closed containers that might be found in the vehicle's passenger compartment.

Belton's basic rationale for both expansions rested not on a concern for officer safety, but rather on an overriding desire to hew "to a straightforward rule, easily applied, and predictably enforced." 453 U. S., at 459.3 When the case was decided, I was persuaded that the important interest in clarity and certainty adequately justified the modest extension of the Chimel rule to permit an officer to examine the interior of a car pursuant to an arrest for a traffic violation. But I took a different view with respect to the search of containers within the car absent probable cause, because I thought "it palpably unreasonable to require the driver of a car to open his briefcase or his luggage for inspection by the officer." Robbins v. California, 453 U. S. 420, 451-452 (1981) (dissenting opinion).4 I remain convinced that this aspect of the Belton opinion was both unnecessary and erroneous. Whether one agrees or disagrees with that view, however, the interest in certainty that supports Belton's bright-line rule surely does not justify an expansion of the rule that only blurs those clear lines. Neither the rule in Chimel nor Belton's modification of that rule would have allowed the search of petitioner's car.

A fair reading of the Belton opinion itself, and of the conflicting cases that gave rise to our grant of certiorari, makes clear that we were not concerned with the situation presented in this case. The Court in Belton noted that the lower courts had discovered Chimel's reaching-distance principle difficult to apply in the context of automobile searches incident to arrest, and that "no straightforward rule ha[d] emerged from the litigated cases." 453 U. S., at 458-459. None of the cases cited by the Court to demonstrate the disarray in the lower courts involved a pedestrian who was in the vicinity, but outside the reaching distance, of his or her car.5 Nor did any of the decisions cited in the petition for a writ of certiorari6 present such a case.7 Thus, Belton was demonstrably concerned only with the narrow but common circumstance of a search occasioned by the arrest of a suspect who was seated in or driving an automobile at the time the law enforcement official approached. Normally, after such an arrest has occurred, the officer's safety is no longer in jeopardy, but he must decide what, if any, search for incriminating evidence he should conduct. Belton provided previously unavailable and therefore necessary guidance for that category of cases.

The bright-line rule crafted in Belton is not needed for cases in which the arrestee is first accosted when he is a pedestrian, because Chimel itself provides all the guidance that is necessary. The only genuine justification for extending Belton to cover such circumstances is the interest in uncovering potentially valuable evidence. In my opinion, that goal must give way to the citizen's constitutionally protected interest in privacy when there is already in place a well-defined rule limiting the permissible scope of a search of an arrested pedestrian. The Chimel rule should provide the same protection to a "recent occupant" of a vehicle as to a recent occupant of a house.

Unwilling to confine the Belton rule to the narrow class of cases it was designed to address, the Court extends Belton's reach without supplying any guidance for the future application of its swollen rule. We are told that officers may search a vehicle incident to arrest "[s]o long as [the] arrestee is the sort of 'recent occupant' of a vehicle such as petitioner was here." Ante, at 8. But we are not told how recent is recent, or how close is close, perhaps because in this case "the record is not clear." 325 F. 3d 189, 196 (CA4 2003). As the Court cautioned in Belton itself, "[w]hen a person cannot know how a court will apply a settled principle to a recurring factual situation, that person cannot know the scope of his constitutional protection, nor can a policeman know the scope of his authority." 453 U. S., at 459-460. Without some limiting principle, I fear that today's decision will contribute to "a massive broadening of the automobile exception," Robbins, 453 U. S., at 452 (Stevens, J., dissenting), when officers have probable cause to arrest an individual but not to search his car.

Accordingly, I respectfully dissent.